J . Am. Chem. SOC.1983, 105, 349-354 than that of C2.H, by at least 13 kcal/mol. The application of such acid-base models to hydrocarbon reactivity is by no means restricted to A g ( l l 0 ) . For example, K o k e previously ~ ~ ~ noted that propylene and toluene could be dissociatively adsorbed on zinc oxide whereas ethylene could not. This behavior was explained in terms of a cutoff in the activity for dissociative adsorption between pK, = 35 and 36 based on aqueous dissociation constants. This behavior is also consistent with the gas-phase acidities of these species; indeed, the acidity difference between C3H6and C2H4is much greater in the gas phase than in aqueous solution. These correlations point out the importance of acid-base properties in catalytic processes such as oxidative dehydrogenation which may involve proton-transfer reactions. Conclusions 1. Propylene reacts facilely with O(a) on Ag( 110) at 140 K to yield OH(a) and C(a). 2. The reaction probability for this
349
reaction is higher than that of ethylene with O(a). A propylene precursor state appears involved. 3. Propylene is less reactive than C2H2which can react to form and displace H 2 0 from the surface at this temperature. 4. Thermal desorption experiments following the reaction of C3H6with O(a) produce H20,C 0 2 ,and C(a). There is no evidence for partially hydrogenated carbon species on the surface a t any temperature, although there is evidence for association of C and 0 in the mixed adlayer leading to CO, formation.
Acknowledgment. We gratefully acknowledge both the National Science Foundation through NSF Eng 23815 and NSF 12964 and the Donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this work. Registry No. Propylene, 115-07-1; silver, 7440-22-4; oxygen, 17778-
80-2.
Theoretical Study of Gas-Phase Methylation and Ethylation by Diazonium Ions and Rationalization of Some Aspects of DNA Reactivity’ George P. Ford* and John D. Scribner* Contribution from the Department of Chemical Carcinogenesis, Pacific Northwest Research Foundation, Seattle, Washington 98104. Received May 18, 1981
Abstract: MNDO semiempirical molecular orbital calculations were carried out for the bimolecular reactions of the methyland ethyldiazonium ions with formamide, imidazole, pyrimidine, methylamine, and water with complete geometry optimization of all stable species and transition states. The activation barriers for methylation were larger, spanned a wider range, and indicated a greater preference for N- rather than 0-alkylation than did those for ethylation. These results were rationalized on the basis of structural and electronic features of the transition states and in turn provided a simple explanation for the parallel behavior observed experimentally when DNA is exposed to diazonium ion precursors and which has been widely discussed in terms of the carcinogenic potency of the latter.
Introduction Many carcinogenic alkylating agents are believed to owe their activity to conversion to highly reactive diazonium ions (1). Among these a r e the N-alkyI-N-nitrosoureas,zaN-alkyl-”nitro-N-nitrosoguanidines,2b1,3,3-trialkyltria~enes,~~ N,N-dialkylnitrosamines,2d 1-aryl-3,3-dialkyltriazene~,~~ and probably the dialkylhydrazines.zf In vivo and in vitro DNA binding studies, in which sufficiently complete product analyses have been carried out, reveal that a great many of the nucleophilic sites are simultaneously alkylated by these agent^.^ However, a clear distinction exists between the methylating and ethylating agents of this type in that the latter react more extensively with the oxygen, relative to the nitrogen centers, than do the f ~ r m e r .This ~ ~~~
~~
(1) Presented in part at the Sanibel Symposium on Quantum Biology and
Quantum Pharmacology, Palm Coast, Fla., March 5-7, 1981. (2) (a) Snyder, J. K.; Stock, L. M. J . Org. Chem. 1980,45, 1990-1999. (b) Lawley, P. D.; Thatcher, C. J. Biochem. J . 1970, 116, 693-707. (c) Sieh, D. H.; Michejda, C. J. J. A m . Chem. Soc. 1981, 103, 442-445. (d) Lai, D. L.; Arcos, J: C. L f e Sci. 1980, 27, 2149-2165. (e) Preussman, R.; von Hodenberg, A.; Hengy, H. Biochem. Pharmacol. 1969, 18, 1-13. ( f ) Preussman, R.; Druckrey, H.; Ivankovic, S . ; von Hodenberg, A. Ann. N . Y. Acad. Sei. 1969, 163, 697-716. (3) See, for example: (a) Beranek, D. T.; Weiss, C. C.; Swenson, D. H. Carcinogenesis 1980, 1 , 595-606. (b) Margison, G. P.; O’Connor, P. J. In “Chemical Carcinogens and DNA”; Grover, P. L., Ed.; CRC Press: Boca Raton, Fla., 1979; Vol. 1, pp 11 1-159. (c) Singer, B. Prog. Nucleic Acid Res. 1975, 15, 219-284.
observation is of special significance in the search for meaningful relationships between chemical structure and biological activity since 06-alkylation of guanine and possibly 04-alkylation of thymine are believed to be critical mutagenic events, whereas 7-alkylation of guanine (the principal site of methylation) may be i n n o c ~ o u s . Thus ~ several workers have sought an explanation for the different reactivities of ethylating and methylating agents of this kind toward the oxygen atoms of nucleic acids and their constituents. Swain-Scott constant^,^ hard-soft acid-base in(4) Singer, B. Nature (London) 1976, 264, 333-339. Singer, B. J . Natl. Cancer Inst. 1979,62, 1329-1339. ( 5 ) For example, the carcinogenic potency of alkylating agents has been shown to correlate with the extent to which they lead to 06but not 7-alkylation of DNA guanine residues6 while the production of tissue-specific tumors correlates with the persistence of 06-alkylguanine in the susceptible tissues.’ It has also been shown that templates containing 06-alkylguanine give rise to transcriptional errors in both DNA and RNA in vitro syntheses, whereas those containing 7-alkylguanine resides do not.* (6) Lawley, P. D. Brit. Med. Bull. 1980, 36, 19-24. Newbold, R. F.; Warren, W.; Medcalf, A. S . C.; Amos, J. Nature (London) 1980, 283, 596-599. Loveless, A. Ibid. 1969, 223, 206-207. (7) Kleihues, P.; Doerjer, G.; Keefer, L. K.; Rice, J. M.; Roller, P. P., Hodgson, R. M. Cancer Res. 1979,39, 5 1 36-5140. Frei, J. V.; Swenson, D. H.; Warren, W.; Lawley, P. D. Biochem. J. 1978, 174, 1031-1044. Goth, R.; Rajewsky, M. F. Proc. Narl. Acad. Sci. U.S.A. 1974, 71, 639-643. (8) O’Connor, P. J.; Saffhill, R.; Margison, G. P. In ‘Environmental Carcinogenesis. Occurrence, Risk Evaluation and Mechanisms”; Emmelot, P., Kriek, E., Eds.; Elsevier/North Holland: Amsterdam, 1979; p 73-96. Singer, B. Prog. Nucl. Acid Res. 1979, 23, 151-194.
0002-7863/83/1505-0349$01.50/00 1983 American Chemical Society
350 J. Am. Chem. Soc., Vol. 105, No. 3, 1983
Ford and Scribner
t e r a c t i ~ n s ,and ~ calculated alkyl group affinities'O have been variously invoked. However, while some of these studies successfully correlate the observations, none satisfactorily explain the underlying physical interactions responsible. We have, therefore, undertaken a theoretical study of the gas-phase reactions of l a and l b with a group of simple oxygen and nitrogen nucleophiles with functional groups characteristic of those in the nucleic acid bases. In this way we hoped to uncover fundamental electronic H2N,
RNjN
m
c=o
HN..&N L3v
H'
2
1
3
c.5
Figure 1. Calculated geometries and (charge distributions) in la and l b .
4
5
6
6 7
AH
AH*-AH'
8
9
10
11 "r eac t i o n c o o r d i n a t e "
a
R=CH3
b
R.C2H5
properties of the species involved which may be of value in the interpretation of the solution-phase reactions of related species, particularly the nucleic acid bases themselves.
Procedure All calculations were carried out by using the M N D O semiempirical molecular orbital procedure described by Dewar and Thiel." For open-shell systems the "half-electron" approximation'2 was used. Where CI was introduced this was based on the eigenvectors of the first excited singlet as described by Dewar and Doubleday." With a few exceptions noted in the text, the geometries of all species were completely optimized with no geometrical constraints. Transition-state geometries were first located approximately from contour maps of the type discussed later. The latter were generated by fixing the forming and breaking bonds at the appropriate lengths (varied in steps of 0.2 A) and minimizing the energy of the system with respect to the remaining variables within the constraints of C, symmetry. Smoothed plots were obtained by quadratic interpolation. Approximate transition-state geometries deduced in this way were refined in a procedure involving the minimization of the scalar gradient of the energy as suggested by McIver and Komornicki.14 Where calculation and diagonalization of the force constant matrix revealed the presence of more than one negative eigenvalue (and the structure therefore not a genuine transition state14), a "down-hill" displacement based on the eigenvectors for each process not connected with the reaction was made and the structure rerefined until the single negative eigenvalue criterion14 was satisfied.
Results and Discussion Methyl and Ethyl Diazonium Ions. The calculated structures and charge distributions of l a and l b are shown in Figure 1. The (9) Lawley, P. D. In 'Chemical Carcinogens"; Searle, C. E., Ed.; ACS Monograph, No. 173, American Chemical Society: Washington D.C., 1976, pp 83-244. Lawley, P. D. In ref 3b, pp 1-36. (10) Pullman, A.; Armbruster, A. M. Theor. Chim. Acta 1977, 45, 249-256. (1 1) Dewar, M. J. S.;Thiel, W. J . Am. Chem. Soc. 1977,99,4899-4907, 4907-4917. Theor. Chim. Acta 1977, 46, 89-104. (12) Dewar, M. J. S.; Hashmall, J. A,; Vernier, C. G. J . Am. Chem. SOC. 1968, 90, 1953-1957. Longuet-Higgins,H. C.; Pople, J. A. Proc. Phys. SOC., London, 19SS,68A, 591-600. (13) Dewar, M. J. S.; Doubleday, C. J . Am. Chem. SOC.1978, 100, 4935-4941. (14) McIver, J. W., Jr.; Komornicki, A. Chem. Phys. Lett., 1971, 10, 303-306. Flanigan, M. C.; Komornicki, A,; McIver, J. W., Jr. In "Modern Theoretical Chemistry"; Segal, G. A., Ed.; Plenum Press: New York, 1977; Vol. 9.
Figure 2. Calculated reaction profile (schematic) corresponding to eq 3 and 4.
M N D O geometry of l a is very similar to that predicted by a b initio calculations at the 4-31GISa and double-{Isb levels. In particular, the present calculations reproduce the interestingly short N-N distance commented upon by Vincent and Radom.lsa The calculated charge distribution is also similar to that predicted by the ab initio calculations's although M N D O underestimates the polarity of the C H groups. The M N D O heat of formation of la (223.5 kcal mol-') is close to an earlierI6 ion cyclotron resonance spectroscopic determination (223 kcal mol-') but higher than a subsequent1' photoionization value (209.4 kcal mol-]). From the latter, and a recent determination of 261.3 kcal mol-l for the heat of formation of CH3', the experimental heat of reaction for eq 1 is 51.9 kcal mol-I, which is almost identical with the ab initio CH3N2' C2HSN2'
-
CH3+ + N2 C2H5'
+ N2
(1) (2)
STO-3G resultLsa(5 1.6 kcal mol-'). Paradoxically, calculations employing superior basis sets lead to markedly lower values. Thus Vincent and RadomlSaobtained A H = 28.5 kcal mol-' at the 4-31G level, while Simonetta and co-workers, using a slightly more flexible basis set designated (9/5), obtained a value of 25.5 kcal mol-l which fell to 18.4 kcal mol-' when polarization functions were added.Isb The corresponding M N D O value is 28.4 kcal mol-'. This unfortunately includes a fairly large, nonsystematic, error in the calculated heat of formation of CH,' which is underestimated by 17.4 kcal mol-I. Using the experimental heat of formation for CH3+leads to a semiempirical prediction of AH = 45.8 kcal mol-' for eq 1. From the calculated heats of formation of C2HS+(219.7 kcal mol-', obsd, 216.0 kcal mol-'I8) and of lb (213.8 kcal mol-'), (15) (a) Vincent, M. A.; Radom, L. J. Am. Chem. Sot. 1978, 100, 3306-3312. (b) Demontis, P.; Ercoli, R.; Gamba, A,; Suffritti, G. B.; Simonetta, M. J . Chem. SOC.,Perkin Trans. 2 1981 488-493. (16) Foster, M. S.; Beauchamp, J. L. J. Am. Chem. SOC.1972, 94, 2425-243 1. (17) Foster, M. S.; Williamson, A. D.; Beauchamp, J. L. Int. J . Mass Spectrom. Ion Phys. 1974, 15, 429-436. (18) Traeger, J. C.; McLoughlin, R. G. J. Am. Chem. SOC.1981, 103, 3647-3652.
J . Am. Chem. SOC.,Vol. 105, No. 3, 1983 351
Gas-Phase Alkylation by Diazonium Ions
Table 1. Calculated Energetic? of the Gas-Phase Bimolecular Alkylation of 2-6 by CH,N,’ (la) and C,H,N,’ ( l b ) methylation ethylation formamide
imidazole pyrimidine me thy lam ine water
Nfb
AH)‘
-39.5 33.2 34.9 -7.5 -60.9
-9.8 -10.2 - 3.9 -5.4 -5.7
~
+
M e
d
17.4 9.8 14.1 14.3 20.9
7.6 -0.4 10.2 8.9 15.2
-47.6 -64.9 -48.3 -45.7 -19.7
~
-9.2 -9.7 -4.8 -3.2 -5.3
*
&-AH’ d
1.8 1.3 7.7 7.5 7.4
11.0 11.0 12.5 10.7 12.7
-44.8 -60.2 -43.6 -41.7 - 16.5
Calculated heat of formation of nucleophile. Other values: AHf(1a) = 223.5; &f(lb) = 213.8; aHf(N,)= All energies in kcal mol‘’. 8.0. C Stabilization energy of ion-dipole complex. AH’= &*(complex) - ZAHf(reactants). Activation barrier relative to reactants. AH* = AHf(transition state) - zMf(reactants). e Heat of reaction.
AH = 13.9 kcal mol-’ is predicted for eq 2. Thus this process is predicted to be -32 kcal mol-’ less endothermic than that of eq 1. The corresponding experimental data are not available. However, some confidence in the M N D O result can be derived from the satisfactory prediction of the analogous quantity for the isoelectronicspecies CH3CO+and C2HSCO+(calcd,’933.8; obsd,20 38.0 kcal mol-’). Even taking into account the likely effects of solvation,2’ the large endothermicity of eq 1 essentially excludes this mode of decomposition of la in aqueous solution at ordinary temperatures while the M N D O results suggest that this cannot be ruled out for l b on similar grounds. However, from a comprehensive survey of the available literature, Friedmanz2 has concluded that there is no experimental basis for invoking free carbenium ions in the reactions of primary alkyldiazonium ions. More recent experimental studies of the alkylation of nucleic acids and their constituents by precursors of n-PrN2+also argue strongly against the significant involvement of the corresponding carbenium ions.23 Gas-Phase SN2Reactions. Next we studied the bimolecular alkylation of the nucleophiles 2-6 (Table I) leading to 7-11 according to eq 3 and 4. The general shapes of the calculated
-
+ CH3N2+ N U + C2HSN2+ Nu
-
+ N2 NuC~HS’ + N2 NuCH3+
(3) (4)
reaction profiles (Figure 2) for the processes described by eq 3 and 4 were similar to those currently accepted for gas-phase ion-molecule reactions.24 In particular, the transition states were preceded by shallow minima corresponding to loose ion-dipole complexes. In each case the nucleophile and the diazonium ion were separated by 3.0-3.4 .f with i the former oriented such that its permanent dipole moment was approximately directed toward the centroid of charge of the latter. The binding in these species was judged to be entirely electrostatic on several criteria: (a) the calculated geometries of the nucleophile and diazonium ion moieties were essentially those of the isolated species (distortions present amounted to
